Training tactical decision making using cognitive models

Simulation-based tactical training can be made more effective by using cognitive software agents to play key roles (e.g. team mates, adversaries). For successful use in training, agents should show tactically representative behavior that support trainees in achieving the learning objectives. For a tactical maritime scenario, we developed such a cognitive agent, playing the enemy force. We used a BDI-architecture to develop a cognitive model of expertise, capable of generating reactive and proactive behavior. This model was implemented in COGNET and then integrated in a training simulation. Evaluation with naive naval officers as subjects showed that the agent successfully supported achievement of the learning goals.